@Test
public void testFixReverseComplementedGenotypes() {
final Allele refA = Allele.create("A", true);
final Allele altC = Allele.create("C", false);
final GenotypesContext originalGenotypes = GenotypesContext.create(3);
originalGenotypes.add(new GenotypeBuilder("homref").alleles(Arrays.asList(refA, refA)).make());
originalGenotypes.add(new GenotypeBuilder("het").alleles(Arrays.asList(refA, altC)).make());
originalGenotypes.add(new GenotypeBuilder("homvar").alleles(Arrays.asList(altC, altC)).make());
final Allele refT = Allele.create("T", true);
final Allele altG = Allele.create("G", false);
final GenotypesContext expectedGenotypes = GenotypesContext.create(3);
expectedGenotypes.add(new GenotypeBuilder("homref").alleles(Arrays.asList(refT, refT)).make());
expectedGenotypes.add(new GenotypeBuilder("het").alleles(Arrays.asList(refT, altG)).make());
expectedGenotypes.add(new GenotypeBuilder("homvar").alleles(Arrays.asList(altG, altG)).make());
final Map<Allele, Allele> reverseComplementAlleleMap = new HashMap<Allele, Allele>(2);
reverseComplementAlleleMap.put(refA, refT);
reverseComplementAlleleMap.put(altC, altG);
final GenotypesContext actualGenotypes =
LiftoverVcf.fixGenotypes(originalGenotypes, reverseComplementAlleleMap);
for (final String sample : Arrays.asList("homref", "het", "homvar")) {
final List<Allele> expected = expectedGenotypes.get(sample).getAlleles();
final List<Allele> actual = actualGenotypes.get(sample).getAlleles();
Assert.assertEquals(expected.get(0), actual.get(0));
Assert.assertEquals(expected.get(1), actual.get(1));
}
}
// necessary to not overload equals for genotypes
private void assertGenotypesAreMostlyEqual(GenotypesContext actual, GenotypesContext expected) {
if (actual == expected) {
return;
}
if (actual == null || expected == null) {
Assert.fail("Maps not equal: expected: " + expected + " and actual: " + actual);
}
if (actual.size() != expected.size()) {
Assert.fail("Maps do not have the same size:" + actual.size() + " != " + expected.size());
}
for (Genotype value : actual) {
Genotype expectedValue = expected.get(value.getSampleName());
Assert.assertEquals(
value.getAlleles(), expectedValue.getAlleles(), "Alleles in Genotype aren't equal");
Assert.assertEquals(value.getGQ(), expectedValue.getGQ(), "GQ values aren't equal");
Assert.assertEquals(
value.hasLikelihoods(),
expectedValue.hasLikelihoods(),
"Either both have likelihoods or both not");
if (value.hasLikelihoods())
Assert.assertEquals(
value.getLikelihoods().getAsVector(),
expectedValue.getLikelihoods().getAsVector(),
"Genotype likelihoods aren't equal");
}
}
public void writeBeagleOutput(
VariantContext preferredVC, VariantContext otherVC, boolean isValidationSite, double prior) {
GenomeLoc currentLoc =
VariantContextUtils.getLocation(getToolkit().getGenomeLocParser(), preferredVC);
StringBuffer beagleOut = new StringBuffer();
String marker = String.format("%s:%d ", currentLoc.getContig(), currentLoc.getStart());
beagleOut.append(marker);
if (markers != null)
markers.append(marker).append("\t").append(Integer.toString(markerCounter++)).append("\t");
for (Allele allele : preferredVC.getAlleles()) {
String bglPrintString;
if (allele.isNoCall() || allele.isNull()) bglPrintString = "-";
else bglPrintString = allele.getBaseString(); // get rid of * in case of reference allele
beagleOut.append(String.format("%s ", bglPrintString));
if (markers != null) markers.append(bglPrintString).append("\t");
}
if (markers != null) markers.append("\n");
GenotypesContext preferredGenotypes = preferredVC.getGenotypes();
GenotypesContext otherGenotypes = goodSite(otherVC) ? otherVC.getGenotypes() : null;
for (String sample : samples) {
boolean isMaleOnChrX = CHECK_IS_MALE_ON_CHR_X && getSample(sample).getGender() == Gender.MALE;
Genotype genotype;
boolean isValidation;
// use sample as key into genotypes structure
if (preferredGenotypes.containsSample(sample)) {
genotype = preferredGenotypes.get(sample);
isValidation = isValidationSite;
} else if (otherGenotypes != null && otherGenotypes.containsSample(sample)) {
genotype = otherGenotypes.get(sample);
isValidation = !isValidationSite;
} else {
// there is magically no genotype for this sample.
throw new StingException(
"Sample "
+ sample
+ " arose with no genotype in variant or validation VCF. This should never happen.");
}
/*
* Use likelihoods if: is validation, prior is negative; or: is not validation, has genotype key
*/
double[] log10Likelihoods = null;
if ((isValidation && prior < 0.0) || genotype.hasLikelihoods()) {
log10Likelihoods = genotype.getLikelihoods().getAsVector();
// see if we need to randomly mask out genotype in this position.
if (GenomeAnalysisEngine.getRandomGenerator().nextDouble() <= insertedNoCallRate) {
// we are masking out this genotype
log10Likelihoods =
isMaleOnChrX ? HAPLOID_FLAT_LOG10_LIKELIHOODS : DIPLOID_FLAT_LOG10_LIKELIHOODS;
}
if (isMaleOnChrX) {
log10Likelihoods[1] = -255; // todo -- warning this is dangerous for multi-allele case
}
}
/** otherwise, use the prior uniformly */
else if (!isValidation && genotype.isCalled() && !genotype.hasLikelihoods()) {
// hack to deal with input VCFs with no genotype likelihoods. Just assume the called
// genotype
// is confident. This is useful for Hapmap and 1KG release VCFs.
double AA = (1.0 - prior) / 2.0;
double AB = (1.0 - prior) / 2.0;
double BB = (1.0 - prior) / 2.0;
if (genotype.isHomRef()) {
AA = prior;
} else if (genotype.isHet()) {
AB = prior;
} else if (genotype.isHomVar()) {
BB = prior;
}
log10Likelihoods = MathUtils.toLog10(new double[] {AA, isMaleOnChrX ? 0.0 : AB, BB});
} else {
log10Likelihoods =
isMaleOnChrX ? HAPLOID_FLAT_LOG10_LIKELIHOODS : DIPLOID_FLAT_LOG10_LIKELIHOODS;
}
writeSampleLikelihoods(beagleOut, preferredVC, log10Likelihoods);
}
beagleWriter.println(beagleOut.toString());
}
public Genotype getGenotype(int ith) {
return genotypes.get(ith);
}